Skip to main content
T0 Big tech Alibaba New grad Round 1

Alibaba AI Infra campus hiring first round

Real Alibaba AI Infra campus hiring first-round interview questions covering inference optimization and related topics

Inference optimization

Fundamentals

  1. Core advantages of Transformer over RNN? Does Transformer also follow a Scaling Law?
  2. Is KV Cache used in training or inference? Why not in training?
  3. Core idea of Flash Attention, and how it optimizes attention compute?
  4. Besides Flash Attention and KV Cache, what other attention optimizations do you know?
  5. Do you know generative recommendation? What is the basic idea?
  6. What is the SID training flow?
  7. Why did Tiger not use a decoder-only architecture?

Project experience

  1. Introduce your internship and main work.
  2. Introduce the overall approach and technical implementation of the news-recommendation project.
  3. In news recommendation, how are positive and negative samples defined?

Coding

  1. Trapping rain water (LeetCode 42).